Fault Diagnosis Of Emulsion Pump Station Based On Data Driving

Emulsion pump station is core equipment of the hydraulic support liquid supply system,also known as power source.Its function is to provide high-pressure emulsion to the working face.Once a failure occurs,it will inevitably affect the support task of mining face.Because it is a complex electromechanical equipmen,harsh working condition,high working intensity,and irregular operation leads to failures frequent and hidden.Long maintenance time affects production efficiency,making the equipment work with diseases,and there is a greater safety hazard.Therefore,it is of great significance for health management of the equipment to characterize mapping relationship between working condition and dynamic monitoring data,and to provide timely early warning,efficient identification,accurate positioning of faults.With revolution of information industry technology,big data and artificial intelligent technology are becoming more and more mature,and the value of data as a mineral has been tapped,which has pointed out direction for fault diagnosis of emulsion pump station based on data driving.Taking BRW500/31.5 emulsion pumping station as research object,combining the traditional fault diagnosis and deep learning fault diagnosis,a hierarchical fault diagnosis method is proposed,and feasibility of the method is verified by experiments.And develop a emulsion pump station fault diagnosis system,verify its function and performance.The main research contents are as follows:(1)Based on the practical experience,combined with the historical monitoring data and related technical information in the remote monitoring of a coal mine,study working performance and form of each component of the emulsion pumping station,sort out structure and common faults of the emulsion pumping station,analysis fault cause,characterize monitoring parameter and analyze failure status of equipment.(2)Aiming at the requirement of condition monitoring of emulsion pumping station,a fault diagnosis method based on parameter diagnosis is proposed.Combined with technical information such as the instruction manual,on basis of long-term data monitoring and analysis,set abnormal thresholds of monitoring parameters and real-time alarms.The data collected by remote monitoring center of the coal mine is selected as data source,and carry out verification experiment.The experimental results show that parameter diagnosis method has a high accuracy of alarming parameters,with it delays 10 seconds,which is practical.(3)To meet the needs of accurate diagnosis of emulsion pumping station,we propose a fault diagnosis method based on deep learning,and construct deep autoencoder network model and deep residual network model respectively.The two methods take eigenvalues of all monitoring parameters at a certain moment and all monitoring parameters at a certain time interval as input,and output fault form end-to-end.Preprocess data collected by coal mine remote monitoring center,and carry out verification experiments.The experimental results show that recognition rates of two methods were 98.712% and 99.212%,respectively.Compared with the latter method,the former method had lower reliability and higher timeliness.It take into account diagnosis accuracy and efficiency,so the diagnosis model of the deep autoencoder network is selected as diagnosis model of the system.(4)Aiming at the needs of fault diagnosis of precision parts of rolling bearings,a fault diagnosis method based on improved densely connected convolutional networks is proposed.It solves the problem that vibration signal is disturbed by environmental noise and weak generalization ability.The Case Western Reserve University bearing data set is selected to carry out verification experiment.The experimental results show that the method achieves99.8%,98.22% and 97.83% in the ideal experiment,anti-noise experiment and variable load experiment,respectively,which has certain advantages.(5)On basis of theoretical research,we develop a fault diagnosis system,which build through Visual Studio 2010 and SQL server 2008R2.It can monitor running condition of epuipment online,analyze fault in real time.The system has functions of condition monitoring,fault diagnosis,data query,etc.After testing,the system runs stably,the diagnosis results are reliable,and easy to use.Fault diagnosis of emulsion pump station based on data driving analyzes the failure mechanism of emulsion pumping station,proposes a hierarchical end-to-end fault diagnosis method combining traditional parameters and deep learning,develops a emulsion pumping station fault diagnosis system.It forms a intelligent mode integrating fault monitoring,diagnosis and disposal of emulsion pumping station,which is helpful for the intelligent health management of emulsion pumping station,ensuring its reliable operation and improving the economic and social benefits of the enterprise.